Electronic Supplementay Materials

contrasting complexity of adjacent habitats influences the strength of cascading predatory effects

James E. Byers1*, Zachary C. Holmes1, Jennafer C. Malek1

1 Odum School of Ecology, University of Georgia, Athens GA 30602 USA. Fax: (706) 542-4819

*Email: ; orcid.org/0000-0001-9240-0287

Appendix S1. Additional methods for, and diagram of, the mesocosm study to examine predation strengths as a function of structure.

Collection methods and husbandry: We acquired oyster spat from a Florida hatchery and stored them for a month in flow-through tanks using unfiltered estuarine water. All other research organisms were collected on-site from the Wilmington River directly adjacent to the mesocosm research site and stored in outdoor flow-through tanks. We collected large mud crabs (Panopeus herbstii) by hand from local oyster reefs and blue crabs (Calinectes sapidus) in recreational crab pots. We fed both crabs a diet of clams (Merceneria merceneria) and mussels (Geukensia demissa). Crabs were starved for 24 hours prior to trials. We collected the bonnethead shark (Sphyrna tiburo) by setting and monitoring a gill net on a nearby oyster reef. The shark was held in a tank identical to those used in the experiment and fed a diet of small blue crabs and store-bought squid once every three days, with feeding days never falling on the day prior to a trial. We allowed for an initial acclimation period of six days after collection before the first trial.

Fig. S1. Split-plot design for trophic cascade mesocosm experiment.

Table S1. Post hoc least squares means tests of generalized linear mixed effects models to examine the effect of trophic treatment on oyster survival within A) sandflat and B) reef habitat in the split-plot mesocosm experiment. The t-test evaluates whether the estimated marginal mean for a given trophic treatment differs from zero, and thus differs from the population level mean. Treatment abbreviations are MC = mud crab, BC = blue crab, and BH = bonnethead shark. Control (i.e. oysters only) and BC+non-lethal BH treatments were used in these comparisons, after coding one oyster in each treatment as dead in order to provide a non-zero variability that could allow statistical calculations. Block nested within trophic treatment was included as a random factor in these analyses.

A) Sandflat

Treatment Least Squares Means /
Treatment / Estimate / Standard Error / DF / t / P /
Oyster / -5.66 / 1.80 / 23 / -3.14 / 0.005
MC / -3.00 / 1.29 / 23 / -2.33 / 0.029
BC / 0.99 / 1.21 / 23 / 0.82 / 0.42
MC+BC / 0.74 / 1.12 / 23 / 0.67 / 0.51
MC+BC+BH / -2.95 / 1.23 / 23 / -2.39 / 0.025
BC+non-lethal BH / -4.52 / 2.27 / 23 / -1.99 / 0.059

B) On-Reef

Treatment Least Squares Means /
Treatment / Estimate / Standard Error / DF / t / P /
Oyster / -5.31 / 1.33 / 23 / -4.00 / 0.0006
MC / 2.16 / 0.80 / 23 / 2.70 / 0.013
BC / -0.52 / 0.77 / 23 / -0.67 / 0.51
MC+BC / 1.48 / 0.72 / 23 / 2.05 / 0.052
MC+BC+BH / 1.75 / 0.72 / 23 / 2.41 / 0.024
BC+non-lethal BH / -4.32 / 1.61 / 23 / -2.69 / 0.013

2